The APL Orchard

One thing I would like to see is a syntax colourer that puts boxes around operators and their operands. Maybe not as far as this but in that direction.

@nathanrogers What picture? Can you show me an example of how it improves APL?

Erik the Outgolfer

@Adám he probably means the pictures over the GH homepage, which are totally unrelated to APL

Adám

@dzaima right, E.g. I wouldn't want ÷∘+/ to show up as ÷∘ +/

nathan rogers

I pasted a link and the readme has pictures

Erik the Outgolfer

yeah, those pictures don't apply to APL

nathan rogers

9:07 PM

except the spacing

which I mentioned specifically

Adám

@nathanrogers Show me an example!

dzaima

@nathanrogers it's not losing universality completely, it's trading a bit for a whole new way to write short functions in the cases where it makes sense. And you decide when it makes sense.

nathan rogers

Erik the Outgolfer

but how is that related to APL?

nathan rogers

the spacing....

Adám

9:09 PM

@nathanrogers Mashing <+> is terrible!

nathan rogers

type in a train and see how it looks

Erik the Outgolfer

except that separate functions should be clearly separate

nathan rogers

but then you'd rather use a train over a non-train....

Adám

I can see mashing ## and :: (which is does) and ⍺⍺ and ⍵⍵ and ∇∇ being useful, but that's it, and only because those are exceptional biglyphs in APL

nathan rogers

functions that are separate should clearly be separate, by explicily passing the argument to the function

Adám

9:11 PM

@nathanrogers So you'd rather write NegSum←{{-⍵}{+/⍵}⍵} than just NegSum←-+/‽

nathan rogers

what's wrong with {(-⍵)+/⍵}

dzaima

@nathanrogers it has many unnecessary things. Also it's wrong :p

nathan rogers

or the minimum amount of things to be direct

why

dzaima

@nathanrogers the ⍵ in {-+/⍵} is pointless and doesn't tell me anything about what the function actually does.

nathan rogers

the ⍵ there is the argument

Adám

9:14 PM

@nathanrogers Wrong because it will evaluate -⍵ and use that as left argument to +/

dzaima

@nathanrogers and explicitly writing it is pointless.

nathan rogers

no it isn't. omitting it is pointless

algebraic notation doesn't omit x's and y's in f(x)'s

dzaima

@nathanrogers it just wastes space & shouldn't get read ever.

@nathanrogers that's.. not an argument.

nathan rogers

yes it is! considering that everything about the notation of APL was designed around algebraic notation!

Adám

@nathanrogers But it does in ±

dzaima

9:15 PM

how is "negate sum of omega" better than "negate sum"?!?

Adám

@nathanrogers Not around; to be better than!

nathan rogers

introtologic.info/AboutLogicsite/…

I mean, this is the entire premise upon with APL's design claims to be based

dzaima

@nathanrogers please inform me of why the ⍵ there is useful and gives me information I couldn't otherwise gather from -+/

nathan rogers

I mean, I'd take it a step further and ask why × exists... because, notation

because to what does -+/ refer to summing?

not a thing

it is a series of symbols that have no meaning without context

dzaima

@nathanrogers it doesn't refer to a specific thing - it lets the caller decide

⍵ doesn't refer to a specific thing either you know

nathan rogers

9:19 PM

but it represents the thing that it will be

which is what variables are

they represent potential values. in trains, those potential values don't exist. where the values belong and in what order is not inferrable without INCREASING THE MENTAL LOAD

dzaima

@nathanrogers you know that tacit programming is a thing, right?

nathan rogers

which as quoted a few times in this converation thus far flies in the face of APLs design philosophy

> By relieving the brain of all unnecessary work, a good notation sets it free to concentrate on more advanced problems, and in effect increases the mental power of the race.

read this phrase
"The horse ran past the barn fell."

Garden-path sentence

A garden-path sentence is a grammatically correct sentence that starts in such a way that a reader's most likely interpretation will be incorrect; the reader is lured into a parse that turns out to be a dead end or yields a clearly unintended meaning. "Garden path" refers to the saying "to be led down [or up] the garden path", meaning to be deceived, tricked, or seduced. Such a sentence leads the reader toward a seemingly familiar meaning that is actually not the one intended. It is a special type of sentence that creates a momentarily ambiguous interpretation because it contains a word or phrase...

trains are the equivalent of this in english

dzaima

The function = alone is read as "equals" not "left arg equals to the right arg", why shouldn't the case be the same with user-defined functions?

nathan rogers

sentences that are technically correct, but require backtracking and reorientation to fully comprehend

that flies in the face of a consistent universal notation that is bent on relieving the brain of all unnecessary work

They're clever, but dumb and bad

equals what?

x = y is read x = y

There is no ability for anyone to infer what is equal without stating the equality!

dzaima

@nathanrogers So why read {-+/⍵}a as "negate sum of omega, called with a"?

nathan rogers

9:23 PM

because that's dumb, write it -+/a

if you need a function f←{-+/⍵} makes sense

dzaima

@nathanrogers And then I move the -+/ out to a variable, and bam, f←-+/

@nathanrogers why change the way to write & read -+/ based on context?

nathan rogers

I'm not negating the value of f g h. I'm arguing fervently about the f g h i j k l alkjasd foaih giowrbunoeurghlakiwehflaiuhryf that it eventaully leads to

you aren't

you're keeping it consistent

-+/ what

-+/thing

dzaima

@nathanrogers and I've been saying all this time that f g h i j k is horrible.

nathan rogers

that's fine

but the existence of f g h enabls aofviuahoieuhewoasdfga

if it didn't exist, neither would the latter

the logical conclusion of universality is that all code is intelligible

the logical conclustion of trains is qoasfhdiauhvpoaBIOVRWOAVNO;IRFJDA;LKRSDJFA;LSEHFiaudH;lJ:LkHLIUGHF;OAEILRJG

dzaima

@nathanrogers 1 2 3 4 5 6 7 8 9 10 11 12 13 also exists, should it not be allowed to make an array of consecutive integers because ⍳13 is better?

@nathanrogers no.

nathan rogers

9:25 PM

AHEM

github.com/DennisMitchell/jellylanguage

dzaima

@nathanrogers how the hell is jelly related?

nathan rogers

we're talking about tacit programming

dzaima

pretty much everyone agrees that Jelly is unreadable.

Erik the Outgolfer

Jelly is a very bad example

nathan rogers

because it has no means of explicitly definiing the positions of arguments!

as I said, it's the logical conclusion of trains

dzaima

9:27 PM

@nathanrogers and it doesn't, because it's trying to be golfy NOT be readable.

nathan rogers

if you remove arguments, and where they belong in an expression, then you remove all intelligbility

dzaima

@nathanrogers disagree.

nathan rogers

trains are attempting to accomplish the same thing

dzaima

disagree.

nathan rogers

I mean, disagree with Ken Iverson all you want

he probably did too by the end of his life

I don't understand how trains leads to anywhere butakvjhabliuhweliauhr;taljie;hotj because that's all I see when I look at them

dzaima

9:29 PM

@nathanrogers that means you're either looking at golfed code, or haven't properly learned them.

nathan rogers

and it flies against the consistency of the notation, leading to further cognative load, along with the Garden Path sentence syndrom I spoke aobut earlier

The old man the boat.

The complex houses married and single soldiers and their families.

Adám

@nathanrogers That's not even true. For years, Iverson struggled to find a neat way to express what in conventional mathematics is written f+g. He experimented with letting every scalar function have a companion operator which combined two functions by applying the corresponding scalar function to their respective results. He added one operator per function by adding an overbar to its symbol. E.g. + had ∓ such that f∓g ⍵ was (f ⍵)+(g ⍵). Finally, on a plane, he got an amazing idea: f+g

nathan rogers

look, english trains

i mean, why didn't he just use +. since there wasn't yet a + that existed which took functions as left and right arguments?

dzaima

@nathanrogers That's the english equivalent of f g h i j k l. I too agree that those are horrible. Not all trains are nice, but some certainly are.

nathan rogers

https://en.wikipedia.org/wiki/Garden-path_sentence

english trains which tacitly refer to the subjects indirectly and through context!

which require mental backtracking, reorientation, and recontextualization

Adám

9:32 PM

@nathanrogers How could as single symbol be both a function and an operator?

nathan rogers

because function and operation is a concept that requires no distinction and was invented in APL for some reason?

what is an operator if not a higher order function?

which we already have a name for

Adám

@nathanrogers Not true. The (traditional) mathematician Oliver Heaviside came up with the terms APL use today.

nathan rogers

It's still an arbitrary distinction

Adám

@nathanrogers Without it, you need special syntax to distinguish function application and passing a function as an argument.

nathan rogers

or consider it a standalone form. + with functions as arguments means something than + with integer arguments, means something different from + with only a right argument

dzaima

9:37 PM

@nathanrogers f g h i j k is horrible. f g h isn't. Sure, there are cases when even that is unreadable, but IMO avg←+/÷≢ is way more readable than avg←{(+/⍵)÷≢⍵}.

nathan rogers

f g h enables f g h i j k

it's existence encourages that kind of thinking

dzaima

@nathanrogers How would you pass functions as arguments? -+÷ 5 is already - (+ (÷ 5))

@nathanrogers NO IT DOESNT

sorry

nathan rogers

it certainly does. as can be seen by any answer that comes through here about just about anythign

Adám

@nathanrogers It ends up very ambiguous: Is ×+÷A equal to (×A)+(÷A) or ×(+(÷A))?

@dzaima You ninja'd me.

dzaima

1 2 3 enables 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20, so should 1 2 3 not be allowed?

nathan rogers

9:39 PM

no because one is necessary

trains are merely additional forms to something that is already expressable

dzaima

@nathanrogers 1 2 3 → 1,2,3

nathan rogers

I can't express 33 723 229748234 247389 274389 42798 234789294837 249387 any other way

dzaima

@nathanrogers 33,723,229748234,247389,274389,42798,234789294837,249387

nathan rogers

yes, but that's nonsense

values in lists are space separated

wut

i can express f g h i j k l m n as a normal expression

which is consistent with the rest of APL

dzaima

@nathanrogers so is {(f ⍵) g (h ⍵) i (h ⍵) j (k ⍵)}, which, tacitly is just f g h i j k (but even that I'd call bad)

Adám

9:41 PM

@nathanrogers No. (1)(2)(3) and 'abc'42'de' have no spaces.

nathan rogers

dzaima

no it isn't

dzaima

@nathanrogers yes it is

nathan rogers

it's minimally direct

dzaima

@nathanrogers f g h i j k is even more direct

nathan rogers

it's the least amount of notation that you can have and specify the operations and their inputs

no it isn't

as noted several times it requires backtracking, and reorientation

it requires first misinterpreting the phrase, then reinterpreting the phrase

dzaima

9:42 PM

@nathanrogers You don't need + to have addition either, it's replacable by a- -b

nathan rogers

and recontextualization

you're just ridiculous

dzaima

@nathanrogers it doesn't for me, if the train is well organized

nathan rogers

these arguments are completely logically inconsistent

Quintec

@nathanrogers Just because you have a hard time interpreting it doesn’t mean everyone does

nathan rogers

a- -b is not conecptually the same as a + b

just because you can interpret it easily doesn't mean that it's good

I can interpret all kinds of broken english

dzaima

9:44 PM

@nathanrogers same with long trains!

nathan rogers

doens't mean that it's grammatically constistent or sound

and short trains

dzaima

@nathanrogers that's where I disagree.

nathan rogers

you're welcome to

my point is that it's disagreeing with the founding tenants of APL from what I can gather in my understanding of APL

dzaima

do you really prefer {(⍺+⍵) , (⍺-⍵) , (⍺ × ⍵) , (⍺ ÷ ⍵)} over +,-,×,÷?

nathan rogers

it's a clever trick, that by merely existing obfuscates otherwise simple code

actually, I would prefer the same thing without parens

but at least the parens are notationally consistent

dzaima

9:45 PM

@nathanrogers but it can't be written without parens

nathan rogers

I know that

dzaima

trains are notationally consistent too you know

and please don't post garden path sentence wiki again

Adám

@nathanrogers You would probably be happy in old-APL land with no train or function compositions. You can write anything without ⍨ or ∘ too.

nathan rogers

But the horse ran past the barn fell. you must not like horses

i don't have a problem with ∘

why would you think that?

dzaima

@nathanrogers a clean, well-spaced, not too long train is as clear as any regular, non-garden path english sentence.

nathan rogers

9:49 PM

and I'm left wondering "what am I adding?? to what? and which am I dividing? rotate what???"

instead of knowing that expressly

mental tacit programming

dzaima

@nathanrogers it doesn't matter in the definition of the function. Knowing that you're adding to x is no better than not knowing what you're adding to

nathan rogers

it's writing code without explicitly reading the arguments

i mean, it really does

dzaima

@nathanrogers you're saying trains are horrible because they're always replaceable with dfns, ∘ and ⍨ are also always replaceable with dfns/parentheses

nathan rogers

Why do we use words. Let's just emoji chat

after all words are just long boring emojis

dzaima

@nathanrogers emoj.li

nathan rogers

9:51 PM

why would I need to read those pesky words

Quintec

it may be frustrating, but with practice it will make sense. Trains follow a set of logical intuitive rules.

nathan rogers

who cares, my emojis mean different things depending on context

dzaima

@nathanrogers the calling context.

nathan rogers

@Quintec it isn't frustrating, it's logically inconsistent. Bending your brain over backwards to justify nonsense leads to frustration. it says so on the box.

Quintec

It’s not logically inconsistent af all.

dzaima

9:52 PM

@nathanrogers "sum divided by tally of 1 2 3" doesn't seem logically inconsistent to me at all

nathan rogers

but it is, as I've mentioned several times, what with the Aurthur Whitehead quote, or the garden path sentences, or the other links I've mentioned

it's very inconsistent with reducing cognative load, plain universal notation without special forms or the need to recontextualize

it really is inconsistent with the values that arise from the design of the language itself

dzaima

@nathanrogers garden path sentences are the equivalent of horribly written trains, how many times do I have to say that

Quintec

That makes no sense. Just because you don’t understand it doesn’t mean it’s inconsistent.

dzaima

^

nathan rogers

just because you understand it doesn't mean that it is

lots of people perfectly understand complete hogwash

they call it Java

Quintec

9:54 PM

Who said that I understand it

dzaima

@nathanrogers java is very, very consistent imo.

Adám

@nathanrogers The values don't arise from the design of the language. The language was designed according to a set of values. Iverson's specifically. Oh, and he came up with the fork notation as a generalisation (like so many other APL features) of the already existing mathematical notation f+g.

nathan rogers

@dzaima fine c++ then

Quintec

Java is as consistent as can be, that’s why it’s so verbose

@nathanrogers same goes^

nathan rogers

@Adám yes I know he came up with it.

c++ is consistent?

read the stl

dzaima

9:57 PM

@nathanrogers can't comment cause I don't really know c++, but it's definitely not nearly as consistent as java

nathan rogers

@Adám that's why I said I'm sure he was arguing with his younger self later on

dzaima

@nathanrogers some things just are nicer when written differently.

nathan rogers

@Quintec look at any 2 C++ code bases, or better yet just look at any highly voted stack overflow article for C++

dzaima

You even earlier said that 33,723,229748234,247389 is nonsense compared to 33 723 229748234 247389, while both mean the exact same thing. Why is the latter better than the former at all?

nathan rogers

@dzaima I'm fine with that, until it leads to more than a fork, then it's nonsense. forks of forks of atops of forks are forking rediculous, and they only exist because the fork exists

because its internally consistent

because the one thing only ever means the one thing

dzaima

9:59 PM

@nathanrogers f g h i j is also consistent

nathan rogers

1,2,3 will always = 1,2,3

no it isn't

dzaima

@nathanrogers (f g h) only ever means (f g h)

Adám

@nathanrogers Actually, J (which was Iverson's attempt at overhauling APL to make mistakes right) started off as explicit by default (like APL) and tacit by request, but Iverson then changed the default to tacit and make explicit requestable. It doesn't seem that he ever regretted that. And that's making tacit the default!

nathan rogers

where are the arguments and in what order are they applied?

Transcript for